CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention

Introduction

This is the PyTorch implementation for the paper titled "CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention".

Dependencies

conda create -n cada_env python=3.10
pip install rl4co

Generate Datasets

Please adhere to the guidelines outlined in the "RouteFinder" repository to create datasets, and then transfer the "data" directory from RouteFinder to this project.

Training

The number of nodes is 50

python run.py experiment=cada-50.yaml

The number of nodes is 100

python run.py experiment=cada-100.yaml